MUPPIT: a method for using proper patterns in model transformations

Model transformation plays an important role in developing software systems using the model-driven engineering paradigm. Examples of applications of model transformation include forward engineering, reverse engineering of code into models, and refactoring. Poor-quality model transformation code is costly and hard to maintain. There is a need to develop techniques and tools that can support transformation engineers in designing high-quality model transformations. The goal of this paper is to present a process, called MUPPIT (method for using proper patterns in model transformations), which can be used by transformation engineers to improve the quality of model transformations by detecting anti-patterns in the transformations and automatically applying pattern solutions. MUPPIT consists of four phases: (1) identifying a transformation anti-pattern, (2) proposing a pattern-solution, (3) applying the pattern-solution, and (4) evaluating the transformation model. MUPPIT takes a transformation design model (TDM), which is a representation of the given transformation, to search for the presence of an anti-pattern of interest. If found, MUPPIT proposes a pattern solution from a catalogue of patterns to the transformation engineer. The application of the pattern solution results in the restructuring of the TDM. While MUPPIT, as a process, is independent of any transformation language and transformation engineering framework, we have implemented an instance of it as a tool using transML and MeTAGeM, which support exogenous transformations using rule-based transformation and OCL-based languages such as ATL and ETL. We evaluate MUPPIT through a number of case studies in which we show how MUPPIT can detect four anti-patterns and propose the corresponding pattern solutions. We also evaluate MUPPIT by collecting a number of metrics to assess the quality of the resulting transformations. The results show that MUPPIT optimizes the transformations by improving reusability, modularity, simplicity, and maintainability, as well as decreasing the complexity. MUPPIT can help transformation engineers to produce high-quality transformations using a pattern-based approach. An immediate future direction would be to experiment with more anti-patterns and pattern solutions. Moreover, we need to implement MUPPIT using other transformation engineering frameworks.